5.4 Seamless Hybrid Integration with Corba 5. Future Research Issues 5.1 Shared Data and Broadcast Issues 4. Hybrid Distributed Computing Capability

Scott Moody is a senior computer science engineer specializing in Computer Language Design and Development. He is a Principle Investigator in the Boeing Defense and Space-Hardware and Software Reuse Initiative. His work deals with researching technologies to support leveraged reuse including: Object Oriented Ada 95 system architectures, distributed processing using Ada 95's Distributed Annex and CORBA, design and development of reusable components, and developing reuse library capabilities including decision modeling and component libraries. Scott is also an active member in the international OMG committee determining requirements for a future Real-Time CORBA capability, and is a member of ACM and the IEEE Computer Society. Scott received a MS and 6 It is conceivable that DSA could support different network transport layers such that communication actually involves CORBA. This was not done for many reasons, and it would not be effective unless the on-the-wire information matched a CORBA specific definition. It would also seem that the idea for configuration files to be first-class (non Ada) objects[5] could be supported without violating any rules. Within this Reuse project, the plan is for configuration files to be automatically generated based on desired products. Some form of hierarchal configuration concept is internally under development for intra-net availability[8]. There are still some fundamental lifecycle differences and costs between CORBA and Ada's Annex-E that must play into an overall development scheme. For example, modification of generated CORBA templates (the IDL specs) requires manual migration into possibly already existing implementation code. The GARLIC system does away with this because the communication code is generated every time changes are made. This is possible since the distribution specification is Ada and not in a neutral language like IDL. Should automatic tools be developed to ease the creation of CORBA IDL specifications from already existing Ada distributed capabilities? The reverse already occurs with existing products-but integration with an application can still be time consuming. It would seem that this tool should map a CORBA model into an appropriate DSA supported model by using the various package categorizations. An Ada-to-IDL mapping should be designed to either automatically or manually support Ada designers as they start to publish CORBA IDL specifications. Since there are different execution issues with the two models, can a correctly running Ada and DSA program be plug replaced with CORBA modules? As a cursory look at the differences, how would priority inversion be solved with CORBA when different task …